Requests

Would you happen to know who invented the term 'data mining'?

Joe Eisenberg IBM Corp.

Note from the editor: I believe that the term "data mining" has first appeared in 1960's in statistics literature. "Data mining" and "data fishing" were used pejoratively to criticize the bad practice of searching the data for correlations without a hypothesis and without proper safeguards against finding patterns due to random noise.

However, with growing volumes of data, data mining began to show potential of finding useful patterns. In late 1980-s, Rakesh Agrawal and other database researchers started to work on association rules, for which they used the term "database mining". However, that term was trademarked by HNC, which had a product "Database Mining Workstation"(tm), and sent strong letters to those who used the term. As a result researchers switched to "data mining".

I came up with the term "knowledge discovery" in March of 1989, to name the first Knowledge Discovery in Databases Workshop. I thought that "data mining" was not sexy enough and "knowledge discovery" sounded better. Little did I know!

Currently, data mining and knowledge discovery are commonly used as synonyms, but data mining is much more popular. Search on Google for the phrase "data mining" gives about 520,000 pages while search for "knowledge discovery" yields about 87,300 pages.

If you have additions or comments on this, please email to editor and I will summarize in the next issue of KDnuggets News.

Gregory Piatetsky-Shapiro
Editor, KDnuggets.